Yeah, that is the issue, isn't it? If PG&E and Anthropic were free to negotiate a plan, I'm pretty sure the agreement would be "In exchange for these large bags of money, PG&E will deliver power next week." That two self-interested/greedy companies don't reach that agreement makes me wonder what's stopping them.

Shortages of people and components are the sorts of problems markets are really good at fixing, maybe not today but in a year or so. If it's not getting resolved, something is keeping the market from working.

Are they not paying? The impression I get is AI companies are expanding as fast as possible (it's a virtual land grab right now) and cost is not the primary concern. Or in other words, if I'm springing for a $10B data center, I can pay $100M to build a transmission line.

If allowed, I'm pretty sure most AI companies would happily pay to build transmission lines and power plants if (a) it got done sooner and (b) they got assurance they'd get the power they need when they need it. They're talking about building their own on-site power plants, fer cryin' out loud, I don't think they're trying to avoid costs.

Dig one level deeper. If selling power is how PG&E makes money, and there's more power to sell if these projects to complete, why is PG&E, being a greedy company, not hiring people in droves to complete them? We've seen other companies, e.g. the AI companies, pay outrageous rates to get good people to complete projects. Why is PG&E (and friends) different?

One obvious difference is the energy distribution companies are regulated monopolies. PG&E, for example, can't just make investments willy nilly--infrastructure spending basically has to get approved by the Public Utilities Commission. They can't decide for themselves what gross profit margin they need to fund investment, that's also controlled by the PUC.

Infrastructure projects also tend to get tied up in the approvals and environmental review quagmire.

I'd be far more likely to believe the issue is heavy handed regulation which is keeping obviously valuable projects from getting completed, far more so than that PG&E is leaving money on the table for reasons.

Oh. I like having a moderate sized screen, about the size of an iPad Mini, for Android Auto and the camera views. My car also has a glass instrument panel which is really nice. It does, however, model physical instruments (speedometer and tach) for ease of reading. It's all the other chrome around the gauges which is nice (next turn, fuel efficiency, performance stats, entertainment details). Is it crucial? Absolutely not. Is it nice? Sure is.

To calibrate you, the car also has a heads-up display for navigation, alerts, and speed. That's one of those "once you try it you're never going back" features. It's so great not having to take my eyes off the road. I'd lose any display save for that one.

I don't generally use the touchscreen controls. It's quite difficult to do that while driving. I'm with you that physical controls are very important for anything you do while the vehicle is moving.

I use Android Auto in my car and that's great for maps and entertainment. It's better the software which comes bundled with the car. The beauty is my car, a BMW, comes with it's fabulous in-console controller, which lets me do virtually everything without moving my hand. Yes, it's a touchscreen, but I almost never use that aspect of it. My impression is other cars have adopted that paradigm.

It still doesn't handle the vehicle controls (heater, AC, sensors). I've got physical HVAC controls, and that's a necessity. All the other sensors (oil level, air pressure, brake pad thickness, etc. etc.) go through the car's software but for as often as I use that, I don't really care what the controls are like.

First, I can't imagine any reasonable need for the FCC to define "broadband" in the first place. What's wrong with letting ISPs and consumers settle on what broadband means?

Second, I have no doubt the goal was to force, jawbone, or subsidize ISPs to deploy 1000/500 network connections. But for the vast majority of people, that's wildly overprovisioned. If all you do is watch AmaFlix and cat videos, you don't even need the 100/20 service which used to be broadband.

So I personally would prefer the FCC would STFU about what is or is not broadband and get back to worrying about radio interference.

I thought the "hedge" in "hedge fund" meant the investment managers were constantly hedging investment bets. In other words, they would have a lot of bets, some of which would only pay off is other ones did not. For example, take some positions which paid off if gold rose, others which paid if gold fell, and now regardless of what gold does, you make some money. The net result is you get a more consistent return rather than what this fund did, spectacular returns followed by a crater. And with that reduced risk comes lower average returns.

What I'd call this fund is a "leverage fund", not a "hedge fund".

When did hedge funds stop hedging bets and become vehicles for highly leveraged, high risk, high return investing? I must have missed that transition.

I have a friend who's pet theory was HR's main goal was to tune compensation to keep the attrition level where the company wanted it. If too few people quit, you're paying too much. If too many, you're not paying enough.

I doubt anyone in HR would put it in quite such cold blooded terms but I think it's at least partly right. Any company needs to think about what type of employees they want and what compensation/benefits/physical environment/work culture will get them that. The right answer isn't always "pay the most for the best of the best".

That seems implausible. Too many disparate organizations have RTO policies for it to be driven by a cabal of real-estate owning moguls.

I can imagine some C-level types saying "we own these building, we have to do something with them." But, since they're flinty hearted greedy bastards (they are, right?), I can more likely imagine them saying "great, we can sell a bunch of real estate and/or terminate our leases and use the proceeds to give ourselves a bonus."

There's an aphorism in encryption development: "anyone can design an encryption system strong enough they personally can't figure out how to break it."

Problem is, of course, that someone else might figure out how to break it. I think we can expect that to eventually happen to any encryption system. It's a matter of when it will happen, not whether. Of course, if "when" is 50 years from now, that's effectively forever.

This is why "I'll write a new encryption system" is almost always the wrong answer. Not always but almost.

There's at least some evidence that AI does not, in fact, lead to job losses. Now, that may be confirmation bias on my part because I'm a free market fanboi but never the less.

Here's the thing though. I'm a software developer in the Santa Clara valley. No one likes layoffs. Everyone knows someone who's been laid off. Virtually every techie I know has been laid off at least once. Unless you're really, really new, you know this is a boom and bust business, and that even the very best engineers at the most stable companies are not 100% secure.

At the same time, virtually every techie I know also has experienced that what our jobs entailed 10, 20, 30 years ago is vastly different from how we work now. Shoot, when I started we still sent paper memos and we shared VT-100 terminals. The only constant is that things always change at a frantic pace. So while AI could be quite disruptive, I'm not ready to confidently panic just yet. It's going to be a jarring transition but so was converting everything from client/server code running on a mainframe to web services and browsers.

Translation: "Give us a bribe and we'll approve your deal."

I know it's difficult to write hard and fast rules but giving politicians unbounded discretion is just asking for corruption.

I'm sure real estate in upstate New York is relatively reasonable. It's mostly still farmland and forests.

Thing is, the people who want to build data centers have no doubt figured that all in. Given the budget for a data center, I'm not sure the land cost is a significant issue. Electricity is something else, which is why people are hot to trot to build a dedicated power plant colocated with the compute.

I'm sorry, your expertise on this comes from...what exactly?

I have no idea how many data centers we need. I've never run the numbers. I wouldn't know what numbers to run. I doubt you have either.

You know who I bet has run the numbers and thinks they can sell a data center's worth of service to willing customers? The person about to write the $10 billion check to build one.

By "there" you mean "the money will remain in California". To an extent, yes. You can live in Austin and found a company where most of the workers are in San Jose. That's just inefficient and a pain. It makes it more likely that, at the margin, a founder will decide Austin is good enough, and hire there.

Can't let that slide. Wealthy people pay the lion's share of income and capital gains taxes, both federally and in California. They pay both more dollars and a higher proportion of income than poor people. The myth that billionaires pay little in taxes is just that, a myth.

FICA, sales, and property taxes may be a different story. There are instances where a rich person pays little but those are the exceptions, not the rule. California in particular is very vulnerable to stock market downturns because capital gains taxes plummet and Sacramento is left with an enormous budget hole.

It gets a bit murky when you start talking about who actually pays corporate income taxes. Some part of it is paid by consumers through higher prices, some by employees through depressed wages, and some by investors through lower dividends and capital gains. I don't think there's general agreement who pays what share of corporate income taxes.